SPECIAL REPORT MEMORY
Making
Connections
The essence of memory is linking one thought to another
by Anthony j. Greene
M
any people wish their
memory worked like a
video recording. How
handy would that be?
Finding your car keys
would simply be a matter of zipping
back to the last time you had them and
hitting “play.” You would never miss an
appointment or forget to pay a bill. You
would remember everyone’s birthday.
You would ace every exam.
Or so you might think. In fact, a
memory like that would snare mostly
useless data and mix them willy-nilly
with the information you really needed.
It would not let you prioritize or create
the links between events that give them
meaning. For the very few people who
have true photographic recall — eidetic
22 Scientific AMericAn Mind
© 2010 Scientific American
memory, in the parlance of the field — it
is more burden than blessing.
For most of us, memory is not like a
video recording— or a notebook, a photo-
graph, a hard drive or any of the other
common storage devices to which it has
been compared. It is much more like a web
of connections between people and things.
Indeed, recent research has shown that
p h O t O i l l U S t r At i O n b Y A A r O n G O O d M A n
some people who lose their memory also
lose the ability to connect things to each
other in their mind. And it is the connec-
tions that let us understand cause and ef-
fect, learn from our mistakes and antici-
pate the future.
The things we remember are the ones
that experience teaches us will help us
make predictions; the newest work in
our laboratory reveals how we make use
j ul y/Au gu s t 2 010
w w w. S c i e nti f i c A m e r i c an .c o m/M in d s c i e n t i f i c am e r i c a n m i n d   23
© 2010 Scientific American
To a mind that can’t make connections, each instant
fleeting and unrelated, each precept without relevance,
of this predictive ability. Other recent
studies show that imagining the future Creating Memories
involves brain processes similar to, but
when a memory forms in the brain, it alters the connections between nerve cells, such
distinct from, those involved in conjur- as these from a sea slug, aplysia. A sensory neuron (left) loses most of its connections,
ing the past. We also tend to remember or synapses, to an adjacent motor neuron when the slug learns that two things or events
the people and events that resonate emo- are no longer related (center). new synapses form when repetition reinforces an associ-
tionally, which is why forgetting an an- ation (right), creating a memory as neurons that fire together wire together.
niversary is such an offense: it is fair evi-
Motor
dence that the date is not as important as neuron
the ones we do remember.
The discovery that memory is all
about connections has revolutionary im- Sensory
plications for education. It means that neuron
memory is integral to thought and that
nothing we learn can stand in isolation;
we sustain new learning only to the de- Ready to go No associations New associations
gree we can relate it to what we already
know. The modern theory of memory er hard drives. Modern scientific dialogue roots in the 1930s and 1940s, when a se-

can help us as we organize our experi-
ences, teach our children and support
those with learning problems.
the history of Memory
For millennia, metaphors for memory
have marched in lockstep with the tech-
nology for recording thought. References
in ancient Greece likened memory to trac-
ings in a wax tablet, in the Middle Ages to
parchment, and later to books, files, pho-
tos, videos, audio recordings and comput-
FAST FACTS
linking and learning
still sometimes refers to memory as a trace
(as in a wax imprint or a rubbing). Lately
we compare human memory to computer
memory, and we draw from the same tax-
onomy. We speak of encoding and storage
for the learning of new memories, retriev-
al for the act of remembering, address for
the location of a memory in the brain, and
output for a remembered event— meta-
phors that persist even as our understand-
ing of memory evolves.
The modern view of memory has its
f r O M a N i N T r o D u C T i o N T o B r a i N a N D B e h aV i o r , t h i r d e d i t i O n , b Y b . kO l b A n d i . Q . w h i S h Aw, w O r t h p U b l i S h e r S , 2 0 0 9
ries of discoveries, most notably by psy-
chologist Karl Lashley of the University
of Chicago and later Harvard University,
revealed that learning and memory are
not sequestered in their own storage
banks but are distributed across the en-
tire cerebral cortex. Lashley set out to
discover the location of the learning cen-
ter of the brain by systematically discon-
necting different regions of the cerebral
cortex in a number of different rats. To
his surprise, all the rats showed some de-
gree of mild learning impairment, but
none was seriously impaired.
The significance of these fi ndings is
profound. It means that memory is dis-
persed, forming in the regions of the

1>> brain cells that fire together during an experience can form perma-
nent connections.
brain responsible for language, vision,
hearing, emotion and other functions. It
means that learning and memory arise

2>> those connections let the same network of brain cells refire later
on, which we experience as memory.
from changes in neurons as they connect
to and communicate with other neurons
[see illustration above]. And it means

3>> new experiences can lead a network of cells to develop further con-
nections, adding to a memory and helping us learn but sometimes
modifying a recollection and creating false memories.
that a small reminder can reactivate a
network of neurons wired together in
the course of registering an event, allow-
ing you to experience the event anew.

4>> because knowledge derives from connections, the optimal strategy

for learning involves making meaningful associations among topics.
Remembering is reliving.
Another piece of the puzzle fell into
place in the 1950s, after some surpris-
ing observations of a few individuals

24 Scientific AMericAn Mind j ul y/Au gu s t 2 010

© 2010 Scientific American
 is an isolated event without continuity, each thought
each person a stranger, every event unexpected.
with almost complete amnesia. In the
most compelling case, a 27-year-old
Connecticut man, known as HM but
identified as Henry Gustav Molaison
after his death in 2008, had severe epi-
lepsy that was not responding well to
medication. It was a sadly common
practice in those days to treat epilepsy
by removing or disconnecting substan-
tial portions of brain tissue. The perfor-
mance of a brain resection on HM re-
sulted in one of the most extreme cases
of amnesia ever recorded. His case and
others revealed that damage to the hip-
pocampus, a wishbone-shaped struc-
ture located deep below the surface folds
of the cerebral cortex, leaves people al-
most completely unable to acquire new
memories or to learn complex associa-
tions. Their minds remain frozen at the
A Neural Switchboard
mat h e w v e r b e r Medical College of Wisconsin
time of the neural insult; the greater the
damage, the more severe the amnesia.
At first, this finding appeared to con-
tradict the discovery that learning and
memory are distributed across the brain.
The hippocampus, though, turned out to
be not the source or storehouse of mem-
ory but rather an essential mediator in
its formation. In a very small brain, ev-
ery neuron might be connected to every
other neuron. But a human brain that
worked on this model would require
that each of hundreds of billions of neu-
rons be linked to every other neuron, an
impossibly unwieldy configuration. The
hippocampus solves this problem by
serving as a kind of neural switchboard,
connecting the distant cortical regions
for language, vision and other abilities
as synaptic networks take shape and
create memories [see illustration below].
Recent amnesia research paints an
even bleaker picture of the condition,
while yielding some startling insights
into what memory really is. Hippocam-
pal amnesiacs appear to have impair-
ments that go well beyond the loss of
memory creation. They also have severe
difficulty imagining future events, living
instead in a fragmented, disconnected re-
ality. In a study published in 2007 psy-
chologist Eleanor Maguire and her re-
search team at University College Lon-
don asked participants with amnesia and
those with normal memories to elaborate
on hypothetical scenarios of short, sim-
ple vignettes. (The researchers gave par-
ticipants a cue card summarizing the
main dramatic elements to ensure that
no one would forget the setup.)
In one scenario, participants were to
imagine they were standing in the main
hall of a museum with many exhibits.
Those with normal memory could gen-
erally develop a coherent narrative about
people and their activities. Amnesic par-
ticipants typically could visualize only a
few details, which were often discon-
nected and lacking in spatial and tempo-
ral coherence. One typical narrative:
“Well, there’s big doors.... There’d be the
exhibits.... I don’t know what they are....
There’d be people.” What is lost in am-
nesia, then, is the capacity to connect
things to one another and find any mean-
ing in them. To a mind that cannot make
connections, each instant is an isolated
event without continuity, each thought
fleeting and unrelated, each precept

without relevance, each person a strang-

er, every event unexpected.

(The Author)
ANTHONY j. GREENE is associate pro-
From its central location, the hippocampus (yellow) connects far-flung regions of the
fessor of psychology at the University
cortex (red) that are involved in a particular recollection. The image, in which the
brain is rendered semitransparent, superimposes a functional MRI image of the of Wisconsin–Milwaukee, where he
cerebral activity on an MRI of its structure. runs a learning and memory lab.

w w w. S c i e nti f i c A m e r i c an .c o m/M in d s c i e n t i f i c am e r i c a n m i n d   25
© 2010 Scientific American
The ability to form and retain connections makes
this critical aspect of learning: “I put two and two

Reimagining Is Reliving
These two fMRIs
show a few differ-
ences but great
overlap in the
regions of the brain
involved in recalling
a past event and
imagining a future
one. In each case,
a network of neu-
rons fires across
the cerebral cortex.

permi ssion from El se vier (brain scans); ge t t y image s (mallard); gk har t/ vikki har t Getty Images (goose); herber t spichtinger Corbis (whistling duck)
and el abor ation,” by Donna Rose Addis, Al ana T. Wong and Daniel L . Sc hac ter, in Neurop sychologia , Vol . 45; 20 07, reprinted with
Remembering the past Imagining the future

The connections across the brain also
help us conceive the future, as recent im-
aging studies have shown. Functional
magnetic resonance imaging detects
changes in blood oxygen in various re-
gions of the brain. When participants per-
form memory tasks, the areas of the brain
that use more oxygen are assumed to be
involved. In 2007 a team led by psycholo-
gist Kathleen McDermott of Washington
University in St. Louis and, separately, by
psychologist Daniel L. Schacter of Har-
vard and his colleagues showed that a mo-
saic of brain areas similar to those in-
volved in memory is active when partici-
pants imagine details of hypothetical or
prospective events. In McDermott’s study,
participants were told they would be
asked to imagine some future event. Once
the subjects were inside the scanner, a re-
searcher would announce a keyword—
“birthday,” say— and the participants
would then think (to themselves; talking
is not permitted during a scan) about how
and where and with whom they would
celebrate a future birthday. Just as memo-
ries do, those imaginings elicited activity
in the hippocampus as well as multiple re-
gions of the cerebral cortex.
In a 2009 experiment with a similar
setup, Schacter’s team asked people to re-
from “Remembering the pas t and imagining the future: Common and di stinc t neur al sub str ate s during e vent con struc tion
call an experience instead of inventing a nections gives us not just a record of
new one. The researchers found that al- events but also the foundations of com-
though the subjects’ brain activity was prehension. We have plenty of clichés to
similar to that of people imagining a new express this critical aspect of learning:
scenario, the patterns were clearly dis- “I put two and two together” or “I just
tinct, suggesting that anticipating the fu- connected the dots.” We make use of
ture is not simply a matter of recasting these connections thousands of times a
past experiences into a novel or hypothet- day. And knowing how they work is cru-
ical form [see illustration above]. Every cial to understanding how we learn.
new event has some novel element to it Connections progress from simple
even if experienced under similar circum- relations between things to ever more
stances, so the pattern of neuronal firing complex cascades of inferences. Links
will be different each time. Our brain has between things, events, people and our
evolved not just for learning and memory actions — so-called item associations —
but for the management of relations: past, are the reason certain objects evoke rem-
present and future. iniscence and become keepsakes. Visit-
ing your alma mater, catching a whiff of
Memory as Connection burning leaves or finding a letter from a
The ability to form and retain con- loved one can bring back vivid memories
The neural connections that make up our memories also help us generalize and discrimi-
nate. Generalization helps us see the common elements in, say, all waterfowl but blinds us
to the differences between the mallard (left), the goose (center) and the white-faced whis-
tling duck (right). In time, we learn to tell one similar creature from another.

26  s c i e n t i f i c am e r i c a n m i n d J u l y/Au gu s t 2 010

© 2010 Scientific American
 comprehension possible. Plenty of clichés express
together” or “I just connected the dots.”
Getting the Most from Your Memory

R
ather than cluttering your mind with arbitrary facts,
try these techniques for using your memory more ef-
fectively. You will understand concepts in greater
depth and have more confidence in what you know.

■ Think about relevant connections between what you

wish to learn and what you already know. Build as many
meaningful connections as you can.
■ Make sure you thoroughly understand what you want
to learn. If you try to memorize a formula, a foreign
phrase or a passage you have not fully grasped, you will
soon forget it and will create a cul-de-sac for future
learning.
■ Explain things to yourself as if you were teaching them
to someone who was unfamiliar with the subject. You
will clarify and consolidate your own understanding.
■ Organize your thinking, outlining complex topics. En- ■ Never cram or rush yourself. Spread out your learning.
sure that the outline is as logical as you can make it. There are no known upper limits on how much human
■ Summarize what you wish to learn so that you establish beings can learn, but there are significant limits on how
the main points. Then elaborate on them to establish much they can learn at the last minute.
depth and breadth. ■ Practice early and often. — A.G.

of events you might not have thought
about in years. Item associations let us
remember that Italy has good wine, and
they help us connect people’s names to
their faces. Items can also be retrieval
cues. A buzzer reminds us that dinner is
still in the oven. The sight of a co-work-
er in the hall prompts us to mark a meet-
ing on the calendar.
Put enough of these item associations
together, and you will create a web of
connections that can help you make pre-
dictions and navigate the world more ef-
fectively over time. At their simplest, pre-
dictive associations remind us that wet
JEFFRE Y C O O LID G E G e t t y I m a g e s
tions requires us to weigh multiple vari-
ables, which in turn takes a brain big
enough to learn all the relations involved.
Indeed, social interactions can pose our
greatest predictive challenges and may
well have been a major impetus, among
our prehuman ancestors, for the evolu-
tion of astounding learning abilities.
At the root of the flexibility of learn-
ing and memory is generalization. My
one-year-old son recently had a won­
derful time feeding some ducks and soon
was able to point out a duck with no
trouble, whatever its color or age. He
also overgeneralizes: in his lexicon, geese
mentary processes that ultimately work
together to shape our associations.
As we amass knowledge over the
course of our life and connect events in
our memory, we learn to model complex
contingencies and make inferences about
novel relations. In my own lab, we have
been exploring the ways people use
learned relations to make predictions. In
one experiment, published in 2006, par-
ticipants lying inside a functional MRI
machine viewed a computer screen on
which we displayed various pairs of un-
familiar shapes, drawn from the Japanese
hiragana alphabet [see top illustration on

snow will clog the snowblower and that
lewd humor will upset some people. But
many predictive associations are more
nuanced. Your boss might get upset if
you tell a crude joke in the office but not
when you are out for a beer and even
then may check to see who else is there
before deciding to laugh. Making predic-
and swans are also ducks. Eventually he
will learn to discriminate among water-
fowl and perhaps in time among differ-
ent types of ducks, as he will later learn
to tell a cabbage from a lettuce and one
sort of lettuce from another. Generaliza-
tion and discrimination are the yin and
yang of learning and memory— comple-
next page]. For each pairing, we asked
participants to click on one shape or the
other, eliciting a message —“correct” or
“incorrect,” depending on the choice —
letting them learn which symbol in a giv-
en pair was preferred. Then we showed
them novel pairings and asked them to
click on the one they thought was correct,

w w w. S c i e nti f i c A m e r i c an .c o m/M in d s c i e n t i f i c am e r i c a n m i n d   27
© 2010 Scientific American
Experiments reveal how easy it is to create false
incorrect information about a car accident, many
rize multiplication tables, the Preamble
Symbol Solutions to the U.S. Constitution, and poetry that
I did not understand at all. Although the
days of learning by repeating things over
and over are finally on the wane, educa-
tors still rely on old-school methods such
as the arbitrary mnemonic device — an
people were trained to recognize which of these symbols from the japanese hiragana
alphabet was the “preferred” one in a series of pairings. then the participants tried to unrelated acronym used as a memory
pick the “correct” symbol in new pairings. the learners who were able to infer a hierar- cue. This is not to say that facts need not
chy of symbols from the original symbol sets made more correct choices later on and be learned. Some, such as phone num-
displayed greater activity in the hippocampus. bers and people’s names, are inherently
arbitrary, and for them rote learning
based on the hierarchy implied during the light was green. Many then remem- may be appropriate. But the message of
the learning phase. Those who scored bered a yellow light that never existed— modern memory research is that the
well— the ones who used inference — dis- which is why eyewitness testimony after brain is wired to recognize and organize
played much greater activity in the hip- police interrogation can be so unreliable. coherent connections, not arbitrary

f r O M “A n f M r i A n A lY S i S O f t h e h U M A n h i p p O c A M p U S : i n f e r e n c e , c O n t e X t, A n d tA S k AwA r e n e S S ,” b Y A . j . G r e e n e , w. l . G r O S S ,
c . l . e l S i n G e r A n d S . M . r AO, i n J o u r N a l o F C o g N i T i V e N e u r o s C i e N C e , v O l . 1 8 ; 2 0 0 6 (t o p) ; w i l l & d e n i M C i n t Y r e C o r b i s (b o t t o m)
pocampus than those who scored nearer To avoid this kind of malleability, smart ones. By tying new learning to existing
the level of chance — the ones who merely lawyers tell clients to write down what associations— by engaging in contextual
guessed— suggesting memory networks happened as soon as possible and before learning— we greatly improve results.
are fully engaged across the brain when discussing it. [For more on the reliability Say, for instance, you were teaching
people use what they know to make pre- of eyewitness testimony, see “Do the Eyes students about a historical novel such as
dictions. This study and others like it Have It?” by Hal Arkowitz and Scott O. The Scarlet Letter. Before you uttered
give insight into how we cobble together Lilienfeld; Scientific American Mind, the name “Hester Prynne,” you might
bits of information learned over many January/February 2010.] fi rst discuss how Puritan society mir-
years and use them to navigate our course rored truths your students already know:
through life. teach, Memory religious leaders do not always live up to
The accretive and adaptive quality of In my own early education during their convictions; the judgment of peers
memory can sometimes cause us prob- the 1970s, schools often taught through is weighty and lasting; concealed shame
lems by altering our memories instead of rote memorization. I was told to memo- eats at the soul. You could next intro-
augmenting them. During the 1990s psy-
chologist Elizabeth Loftus, now at the
University of California, Irvine, pro-
duced an impressive amount of research
showing how easy it is to create false
memories of past events. In one study,
participants watched a film of a car acci-
dent. Researchers asked some subjects
how fast they thought the cars were going
when they “smashed into” each other
and asked other subjects how fast the cars
were going when they “hit” each other.
The subjects who heard the word
“smashed” gave significantly higher esti-
mates of the speed. In other experiments,
subjects were fed incorrect information
about an accident after watching the film;
they might, for instance, be asked repeat-
edly whether a traffic light had turned new brain insights suggest that rote learning is usually ineffective. Students are more likely
yellow before the collision when in fact to retain new information when they can relate it to what they already know.

28 Scientific AMericAn Mind j ul y/Au gu s t 2 010

© 2010 Scientific American
 memories of past events. When people were fed
remembered a yellow light that never existed.
duce ways in which the Puritan universe
was different: living on the verge of sur-
vival necessitated collective conformity;
technology was primitive; religious con-
victions were a community affair. Your
students can then engage their imagina-
tions to paint a picture of the characters’
way of life, from family relations to dogs
chasing pigs down the streets. As their
knowledge of the story’s setting deep-
ens, they can begin to take in abstract
ideas, such as the political and legal
structures of Puritan society. By this
time, the students will have formed an
intricate web of associations that will let
them weave the lessons of the book into
their own thinking. They can live the Rather than learning multiplication tables by endless repetition, students should figure
out how many wheels they would need to equip a fleet of cars.
story and grasp its significance.
Contextual learning can even help
with tasks that seem intractably a mat- from another, and from there exceptions Memory springs from the connections.
ter of rote memorization, such as learn- to those conditions, and in time more Despite our imperfect metaphors,
ing the times tables. You can better teach complicated relations still. we must have long known that memory
a child that 3 times 4 equals 12 by bring- Connections help us anchor an ever was more than a mere repository of ex-
ing the idea into the real world. Ask a more complicated body of knowledge perience. Consider the ancient Greeks.
girl who likes cars how many wheels she about how the world works and negoti- The muses were not simply the goddess-
would need for three racers with four ate the complex structures all around us. es who inspired poetry, music and all
wheels apiece. As she memorizes the ta- Memory is a dynamic aspect of our in- forms of artistic creativity and for whom
bles, she will learn what multiplication tellect. And as our understanding of temples were erected. They were also
actually does, which will help her solve memory grows deeper, we see that the patrons of the liberal arts and the well-
related problems later on. When rich in connections we make between the peo- spring of philosophy, knowledge,
context, science becomes an extension ple, places and things in our lives, be- thought and wisdom. And the mother
of a student’s natural curiosity to figure tween the past, present and future, do and queen of all the muses? Mnemos-
out how things work, just as history, as not themselves spring from memory. yne, the muse of memory. M
collective memory, connects students to
the continuum of civilization, and fine (Further Reading)
arts connects them to the mind of the
artist. If the connections between sub- ◆ ◆ Metaphors of Memory: A History of Ideas about the Mind. Douwe Draaisma.
Translated by Paul Vincent. Cambridge University Press, 2000.
ject matter and students are relevant and
◆ ◆ Contextual Teaching and Learning: What It Is and Why It’s Here to Stay. Elaine B.
personal to them, the learned material Johnson. Corwin Press, 2002.
becomes part of their beings. ◆ ◆ Synaptic Self: How Our Brains Become Who We Are. Joseph LeDoux. Penguin
The lessons of associative learning Books, 2002.
JEFFRE Y C O O LID G E G e t t y I m a g e s

can also give educators a fresh approach
to early learning and to teaching children
with learning deficits. Simplified contin-
gencies can help everyone in the first
stages of learning. For example, a young
child who thinks that all swimming birds
are ducks may later learn conditions that
distinguish one kind of swimming bird
◆ ◆ What’s New with Amnesic Patient HM? Suzanne Corkin in Nature Reviews Neuro-
science, Vol. 3, pages 153–160; 2002.
◆ ◆ From Conditioning to Conscious Recollection: Memory Systems of the Brain.
Howard Eichenbaum and Neal J. Cohen. Oxford University Press, 2004.
◆ ◆ The Science of False Memory. C. J. Brainerd and V. F. Reyna. Oxford University
Press, 2005.
◆ ◆ The Hippocampus Book. Edited by Per Andersen, Richard Morris, David Amaral,
Tim Bliss and John O’Keefe. Oxford University Press, 2007.

w w w. S c i e nti f i c A m e r i c an .c o m/M in d s c i e n t i f i c am e r i c a n m i n d   29
© 2010 Scientific American